gchristensen changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html | 18.03 release managers: fpletz and vcunat
hl has quit [Ping timeout: 240 seconds]
primeos[m] has quit [Ping timeout: 245 seconds]
peterhoeg has quit [Ping timeout: 256 seconds]
olejorgenb[m] has quit [Ping timeout: 256 seconds]
grahamc has quit [Ping timeout: 246 seconds]
codyopel has quit [Ping timeout: 246 seconds]
adisbladis[m] has quit [Ping timeout: 240 seconds]
regnat[m] has quit [Ping timeout: 240 seconds]
thefloweringash[ has quit [Ping timeout: 240 seconds]
rycee has quit [Ping timeout: 240 seconds]
hedning[m] has quit [Ping timeout: 240 seconds]
mbrgm has quit [Ping timeout: 264 seconds]
florianjacob has quit [Ping timeout: 246 seconds]
copumpkin has quit [Ping timeout: 256 seconds]
FRidh[m] has quit [Ping timeout: 256 seconds]
sphalerit has quit [Ping timeout: 256 seconds]
pstn has quit [Ping timeout: 276 seconds]
dtz has quit [Ping timeout: 255 seconds]
mbrgm has joined #nixos-dev
<gchristensen> Dezgeg: any chance you're around?
alp has quit [Ping timeout: 240 seconds]
alp has joined #nixos-dev
<Dezgeg> for a bit probably
<gchristensen> Dezgeg: may I PM? I have some "off topic for every channel" questions
<Dezgeg> go ahead
kmicklas has quit [Ping timeout: 264 seconds]
orivej has quit [Ping timeout: 240 seconds]
<samueldr> hope niksnut looks at the IRC backlog (I may ask tomorrow too) hi! you can ignore the [big!] PR if you don't have the time for it; but unless there's anything wrong with them, I have three simple fixes with big impact for nixos-homepage https://github.com/NixOS/nixos-homepage/pulls/samueldr thanks!
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm has joined #nixos-dev
Mic92 has quit [Ping timeout: 260 seconds]
goibhniu has quit [Ping timeout: 265 seconds]
Mic92 has joined #nixos-dev
mbrgm has quit [Ping timeout: 246 seconds]
mbrgm has joined #nixos-dev
Lisanna has quit [Quit: Lisanna]
pie_ has joined #nixos-dev
Lisanna has joined #nixos-dev
mbrgm has quit [Ping timeout: 240 seconds]
mbrgm has joined #nixos-dev
<Lisanna> samueldr that's an awesome improvement!
obadz- has joined #nixos-dev
sphalerite_ has joined #nixos-dev
obadz has quit [Ping timeout: 266 seconds]
sphalerite has quit [Ping timeout: 266 seconds]
obadz- is now known as obadz
Lisanna has quit [Quit: Lisanna]
grahamc has joined #nixos-dev
ma27 has joined #nixos-dev
pie_ has quit [Write error: Connection reset by peer]
primeos[m] has joined #nixos-dev
codyopel has joined #nixos-dev
peterhoeg has joined #nixos-dev
regnat[m] has joined #nixos-dev
hedning[m] has joined #nixos-dev
olejorgenb[m] has joined #nixos-dev
adisbladis[m] has joined #nixos-dev
sphalerit has joined #nixos-dev
dtz has joined #nixos-dev
thefloweringash[ has joined #nixos-dev
pstn has joined #nixos-dev
rycee has joined #nixos-dev
copumpkin has joined #nixos-dev
florianjacob has joined #nixos-dev
hl has joined #nixos-dev
FRidh[m] has joined #nixos-dev
ma27 has quit [Ping timeout: 260 seconds]
sphalerite_ is now known as sphalerite
jtojnar has joined #nixos-dev
<viric> :q
<viric> oops.
<LnL> I can't get impureEnvVars to work anymore
pie_ has joined #nixos-dev
tilpner has quit [Quit: :wq]
tilpner has joined #nixos-dev
ma27 has joined #nixos-dev
<Mic92> LnL: https://hydra.nixos.org/build/72883050/nixlog/1 looks like there are more flacky tests to kill on macOS
<LnL> have you seen that more then once?
pie_ has quit [Ping timeout: 260 seconds]
sbu has joined #nixos-dev
sbu has quit [Client Quit]
__Sander__ has joined #nixos-dev
pie_ has joined #nixos-dev
ckauhaus has joined #nixos-dev
pie_ has quit [Quit: Leaving]
orivej has joined #nixos-dev
qqlq has quit [Ping timeout: 268 seconds]
zybell has quit [Quit: Leaving]
zybell has joined #nixos-dev
zybell has quit [Read error: Connection reset by peer]
zybell has joined #nixos-dev
qqlq has joined #nixos-dev
tilpner has quit [Quit: :wq]
tilpner has joined #nixos-dev
<domenkozar> what's the state with musl? does it yet work with haskell packages to provide a statically linked executable?
Synthetica has joined #nixos-dev
<clever> domenkozar: its using musl for the namespace helper, but not for the haskell arx that generates things
<Profpatsch> niksnut: Do you still remember the reasoning for structuring escaping inside of '' the way you did?
<Profpatsch> ''$ and '''
<Profpatsch> It’s so nice to use in practice, compared to backslash escaping.
<Profpatsch> But I don’t really see why.
<Profpatsch> And maybe the orgiginal reason for doing it that way helps me understand.
<Profpatsch> I feel it has something to do with ' already being the delimiting symbol versus using another symbol like \ for escaping.
<niksnut> it had to do with minimizing the amount of escaping necessary in particular for shell code
<shlevy> In retrospect I kind of wish we didn't use ${} for nix interpolation for that reason
<Profpatsch> shlevy: I object, I kind of wish we didn’t use bash for our model of execution. :)
<samueldr> [appropriate time-dependent greeting] I think I have a fix for the options page not updating too
<shlevy> Profpatsch: guix exists ;)
<Profpatsch> niksnut: Ah, to use a different escaping so that nix and bash escaping don’t interact. Makes sense.
<Profpatsch> But even interpolating nix strings inside of nix multiline strings is kind of pleasant.
<Profpatsch> Event interploating bash strings inside of nix expressions inside of nix multiline strings to be more specific. :D
ma27 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos-dev
<domenkozar> clever: I'd want just single a single executable statically linked
<shlevy> dtz: ^ :)
<domenkozar> obviously a single single, not double single :-)
<clever> domenkozar: i'm not sure why nix-bundle avoided musl for arx, its either broken, or he was just avoiding a ghc rebuild for a build-time only tool, where binary size didnt matter
<domenkozar> I think musl wasn't near ready at least a month ago when I looked, but I know there was work done on this front
<niksnut> in retrospect I think something like <<...>> would have been nicer
<gchristensen> cmd >> outfile
* shlevy shakes his fist at bravo
<niksnut> blame it on bravo ;-)
<niksnut> actually we should have gone unicode, e.g. «…» or “…”
<gchristensen> now we're cooking
<niksnut> or ┏ … ┛
<gchristensen> this would at least end the complaints about docbook
<gchristensen> what if we replace '' ... '' with '​' ... '​' (that is to say, a single quote, zero width space, single quote)
<zybell> it would start complaints: How do I type ... ? instead;-)
<niksnut> I like your thinking
<samueldr> how do you then quote those inside? the solution is '[zw]+' with the same balanced amount of zero-width spaces each side!
<gchristensen> samueldr: did you see the ruby proposal to collapse `end end end` with `ennnd`?
<zybell> There is something like that in pod (but more visible)
<samueldr> I'm sorry, I know you're (probably) busy, niksnut, but for those other three PRs on nixos-homepage, if there's feedback and changes, I'd like to know, thanks! https://github.com/NixOS/nixos-homepage/pulls?utf8=✓&q=is%3Apr+is%3Aopen+LGTM
<samueldr> (and thanks for merging the other one earlier)
<Profpatsch> Isabelle uses the cartouche « » as a symbol that is outside of the alphabet used by the languages embedded in it.
<Mic92> Security stuff is still commited directly to master? https://github.com/NixOS/nixpkgs/pull/39045
<domenkozar> -buffer #servant
<domenkozar> meh
<gchristensen> Mic92: looking
<Mic92> gchristensen: maybe it is not worth if the bug is not that critical.
<Mic92> it is only a NULL-pointer dereference
<gchristensen> Mic92: yeah, agreed
qqlq has quit [Ping timeout: 240 seconds]
qqlq has joined #nixos-dev
<LnL> lol, tar: tools: time stamp 2018-04-17 11:48:19 is 1786.481265469 s in the future
pie_ has joined #nixos-dev
<Profpatsch> Another very interesting property of '' … '' is that the delimiters are not different.
<gchristensen> is that very interesting?
<Profpatsch> gchristensen: It feels very interesting at least.
goibhniu has joined #nixos-dev
<gchristensen> aren't most delimiters the same?
<Profpatsch> But it could be that it’s not very interesting in a more theoretical sense.
<Profpatsch> Or even for a parser really, because it just changes what you look for in your current context.
phreedom has quit [Ping timeout: 268 seconds]
<ckauhaus> does anyone know whom to bug to review a NixOS installation manual PR?
<ckauhaus> not sure who is responsible
<gchristensen> I'm not sure anyone is responsible
<ckauhaus> heh
<Profpatsch> heh
<makefu> ckauhaus: you think it is a good idea to remove the example setup with pvcreate,lvcreate etc? because i am very sure not all users are 100% fond with the exact usage
<ckauhaus> yes, this is exactly the problem
<ckauhaus> when you type it as is it won't work
<ckauhaus> so in my opinion it is better to leave out detailed but misleading instructions
<ckauhaus> or do it right
<makefu> at least we could provide a minimal working example
<ckauhaus> makefu: would be nice of course, but in this case the text should be restructured
<ckauhaus> there is no point in creating an ext4 volume on the bare partition first and then slapping a PV on it
<ckauhaus> this will result in exactly the error the OP puts out in the ticket as second point
<makefu> true
<ckauhaus> so I think it's better to go along the line with mdadm mentioned below in the same text and only point out that these tools might be of use
<ckauhaus> there are really plenty of LVM intros out in the 'net
<makefu> mh,mh i see what you mean
<makefu> the best thing then would be to have a minimal working setup with for example the boot partition and ext4 as root
<gchristensen> I'd rather not push people towards LVM anywho. it is complicated to get it right, and is complicated to recover from when there are prodbems
<ckauhaus> yeah
<makefu> i think so as well, especially on single-user setups such as a laptop lvm can be overkill
<zybell> ckauhaus:You could *link* to a LVM example from the LVM commands.
<ckauhaus> yes
<ckauhaus> well
<ckauhaus> perhaps
<ckauhaus> what about mdadm?
<ckauhaus> I think the scope of the installation manual is to provide steps to get NixOS running
<ckauhaus> this is not a generic Linux admin primer
<gchristensen> +1
<ckauhaus> but feel free to submit a PR - I'm not the owner of that document ;-)
* gchristensen looks at the last committer
<gchristensen> you are now! ;)
<ckauhaus> uhm well
<zybell> The problem is to get it running in a way that leaves options for later.
<gchristensen> someone who needs that information will find it
<ckauhaus> they keep it very basic as well
<zybell> especially if some ....enable=true would initiate a complete reinstall I would be very surprised!
<ckauhaus> please go ahead, improve the text and come up with a better solution :)
thefloweringash has joined #nixos-dev
phreedom has joined #nixos-dev
<zybell> did you look in the wiki? They do it with linking. Did you follow the link to LVM? Complete Description incl go back to Install link at the right place!
* zybell is impressed.
<xCuri0> How do I port NixOS to a new architecture ?
phreedom has quit [Remote host closed the connection]
<gchristensen> xCuri0: I can't help, but what architecture?
<xCuri0> gchristensen, arm64ilp32
<xCuri0> its already supported by linux and debian has a port of it
phreedom has joined #nixos-dev
<gchristensen> what is ilp32?
kmicklas has joined #nixos-dev
<xCuri0> gchristensen, 64-bit using 32-bit pointers which reduces memory usage
<xCuri0> which increases performance too but limits memory to 4gb (which is ok on 4gb or less systems)
<LnL> git show bbbbbbc
<LnL> :o
<gchristensen> :D
<Mic92> 32bit pointers in 64bit allows to create Spectre mitigation.
<Mic92> by limiting user space to 32bit while kernel pages reside in the 64bit one.
<LnL> heh, I was going to se we don't have a deadbeef yet :p
<LnL> say*
<zybell> Mic92:That mitigates Meltdown,Spectre is about a user prog compromising *itself* by speculative exec.
<Mic92> zybell: reading your own memory is borring
<Mic92> reading kernel pages in the same page table is much more compelling.
<Dezgeg> but you can still try to access the memory above the 64-bit boundary, so I don't see how it mitigates anything
<Mic92> not if you cannot address those pages.
<Dezgeg> well in x32 and arm64ilp32 you can
<Mic92> yes. but this could be changed.
<Dezgeg> how?
<Dezgeg> it's just the processor executing regular 64-bit code
<Dezgeg> just with the self-imposed limitation that no userspace pointer is over 32-bit
<Mic92> ah the paper was for “meltdown” not spectre
<xCuri0> isn't spectre patched already
<xCuri0> and i think amd's new ryzen 2 has the silicon patched
<xCuri0> according to what they said earlier
<Mic92> there is still old hardware though
<xCuri0> doesn't the kernel have mitigations too even if the microcode doesn't get an update ?
orivej has quit [Ping timeout: 264 seconds]
<clever> Dezgeg: this reminds me about https://github.com/raspberrypi/firmware/issues/550
<zybell> Mic92:A javascript from darkweb in firefox reading the memory of password cache in firefox is boring? I tend to disagree!
<clever> Dezgeg: the issue they are discussing there, is that the GPU blob uses 32bit opaque fields, to pass hold userland pointers in gpu memory, and pass them back to userland at a later point in time
<clever> Dezgeg: but with a 64bit userland, the pointers dont fit!
gleber_ has quit [Ping timeout: 276 seconds]
<clever> s/pass//
<xCuri0> how do i (try) to compile NixOS for arm64ilp32 ?
<xCuri0> whoever does the porting to arm should now
<Dezgeg> look up the commit that introduced e.g. risc-v and imitate that
<xCuri0> Dezgeg, in which repo ?
<xCuri0> nix ?
<Dezgeg> nixpkgs
<xCuri0> i see 16 commits about adding risc-v support to different packages
<xCuri0> ill try to replicate those for ilp32
<Dezgeg> probably no package needs to be changed
<Dezgeg> hm, I don't actually know how this risc-v port even works since it's missing some stuff I thought it'd have
<gchristensen> it seems I can't use nix-build if I don't have internet ...
<Dezgeg> I guess you can try to imitate aarch64 then which added things like else if targetPlatform.system == "aarch64-linux" then "${libc_lib}/lib/ld-linux-aarch64.so.1" to various places
<Dezgeg> but the commit that added aarch64 is much older so some things have probably changed
<xCuri0> once i edit it how do i build it ?
<xCuri0> do i manually build each package ?
<clever> xCuri0: nix will automatically build the entire closure, so you just need to ask for hello world, and watch it run
<clever> xCuri0: but, you need to decide on native or cross-compiling first
<xCuri0> clever, so i just tell it to use my nixpkgs repo instead of the default and configure it for my custom arch ?
<clever> xCuri0: yeah
<xCuri0> what about building ?
<xCuri0> native build would take too long probably
<clever> if you are doing native builds, you need an existing arm64ilp32 build slave with nix installed
<clever> if you choose to cross-compile, you have to set the right flags to nixpkgs to tell it to cross-compile
<Dezgeg> pkgs/stdenv/linux/make-bootstrap-tools-cross.nix is a good starting point which already has the flags set
<jtojnar> gchristensen: yes, it is pretty annoying, I had to use --option substituters ''
<gchristensen> cc shlevy
<clever> Dezgeg: i believe that one is only needed for bootstrapping native builds
<Dezgeg> well yes, but it's a starting point still
<xCuri0> anyone know how big nixpkgs is before i clone it ?
<gchristensen> ~1.4g
<clever> mine is 900mb due to zfs compression
<viric> clever: how do you know the compressed size?
<clever> viric: du defaults to the on-disk size, rather then the raw sum of how many bytes are in the files
<clever> so for zfs, that will be less then the real size
<clever> and for all other fs's (and zfs without compression), more, due to the overhead of cluster size
<viric> btrfs intentionally did not do that
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
<clever> 4.5K -rw-r--r-- 1 clever users 234 Apr 17 10:59 default.nix
<viric> because that broke some userland tools.
<clever> for this file, compression doesnt help, its too small
<clever> 389K -rw-r--r-- 1 clever users 1.9M Apr 17 05:21 default.nix
<viric> I ask because btrfs people decided not to go that way
<clever> but for this one, it helps a lot
<viric> (which for me was the obvious way)
<viric> they have a separated syscall for compressed size.
kmicklas has quit [Ping timeout: 256 seconds]
<clever> Size: 1888254 Blocks: 777 IO Block: 131072 regular file
<clever> 389K -rw-r--r-- 1 clever users 1.9M Apr 17 05:21 /home/clever/iohk/cardano-develop/pkgs/default.nix
<clever> viric: oh, and i have no idea how the values in stat convert to what ls displayed
<viric> 512 block size, usually
<clever> ah
<clever> that does match up
orivej has joined #nixos-dev
<Sonarpulse> Dezgeg: ping
<Dezgeg> pong
<Sonarpulse> and I pinged you in a few other PRs too
<Sonarpulse> (also sorry I alt-tabbed, if you `sonarpulse: pong` I'll hop back instantly(
<Dezgeg> I saw it, just been a bit busy
<Sonarpulse> Dezgeg: sure
<Sonarpulse> I guess that and the general system localSystem platform stuff
<Sonarpulse> I'd like to move forward writing the docs or whatever the next step is
<Sonarpulse> I'm happy to do the work just need some agreement on direction first
__Sander__ has quit [Quit: Konversation terminated!]
<Sonarpulse> Dezgeg: like if you are too busy to look over, maybe comment like "I guess this fine for now as long as you don't add more enableIfSupported"
<Dezgeg> generally I feel this lib.enableIfAvailable sacrifices readability
<Sonarpulse> Dezgeg: well that's what I mean putting that on freeze
<Sonarpulse> like I think the allowBroken change is perhaps a good idea regardless?
<Sonarpulse> I guess I'm just trying not to be blocked
<Sonarpulse> if there's some way I can keep working without you needing to decide everything now, that would be great
<Dezgeg> the allowBroken change is fine but the fact that it causes really unpredictable changes when enableIfSupported seems totally confusing
<Dezgeg> I mean separating allowBroken and allowUnsupportedSystem
<xCuri0> what toolchain does nixos use for risc-v ?
<xCuri0> can't find anything about the toolchain in the commits
<sorear> what do you mean by the question
<sorear> there's only one toolchain that works, gcc
<Profpatsch> Do we have a solution to the “one fixpoint to bind them all”-problem?
<Sonarpulse> Dezgeg: great! well if you can comment as much on that PR
<Sonarpulse> I'll do the docs and merge it
<Profpatsch> As in integrate haskellPackages, pythonPackages … into the main fixpoint?
<Sonarpulse> and we'll have a "stay" on further usage of enableIfSupported until you way in
<Sonarpulse> Dezgeg: to be clear, I don't want to prevent you from having a say because you are busy at this moment :)
<xCuri0> sorear, normally for building for other archs you do something like aarch64-linux-gnu-gcc
orivej has quit [Ping timeout: 256 seconds]
<xCuri0> is the risc-v port finished since looking at the code it seems half done ?
<gchristensen> definitely not
<xCuri0> this seems more useful than searching commits https://github.com/NixOS/nixpkgs/pull/35092
<xCuri0> well gtg bye
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
ma27 has quit [Ping timeout: 256 seconds]
ma27 has joined #nixos-dev
ma27 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos-dev
ma27 has quit [Client Quit]
ma27 has joined #nixos-dev
Lisanna has joined #nixos-dev
gleber_ has joined #nixos-dev
Synthetica has quit [Quit: Connection closed for inactivity]
orivej has joined #nixos-dev
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-dev
taktoa has quit [Remote host closed the connection]
<Sonarpulse> shlevy: really quick, what if i renamed unsupported systems -> unintended systems?
<Sonarpulse> I worry "supported" is too strong
<Sonarpulse> that would be hydraPlatforms if anything
<gchristensen> supported is interesting because it isn't necessarily saying upstream supports it or not, but possibly if _we_ support it or not
<Sonarpulse> gchristensen: yeah and I think that's the wrong thing actually imo
<Sonarpulse> meta.platfroms I think should closely adhere to upstream
<Sonarpulse> (maybe baring windows or something crazy :D)
<Sonarpulse> and then meta.broken can blacklist things upstream support that we don't
<Sonarpulse> signaling the todo
<Sonarpulse> also, if its allowed by meta.platforms, but not built by hydra (for whatever the reason), is it really *supported* on that platform?
<Sonarpulse> or the lesser tier of "not declared known broken"
<Sonarpulse> I don't want to dilute "supported" for half-maintained things
<gchristensen> I'm not sure that is true
<gchristensen> because in a sense, who cares what upstream declares, it is our packaging of it
<shlevy> Sonarpulse: I'm not sure about the best name there really.. unintended seems off but I see your concern with supported
<Sonarpulse> gchristensen: now that I wrote both, I think my "dilution", "boy who cried ~wolf~ supported" point is the more important one :)
<Sonarpulse> shlevy: in any event, I won't bog down your PR with this :)
ckauhaus has quit [Quit: Leaving.]
Jackneill has quit [Read error: Connection reset by peer]
Jackneill has joined #nixos-dev
ma27 has quit [Ping timeout: 240 seconds]
zybell has quit [Ping timeout: 256 seconds]
zybell has joined #nixos-dev
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev